The statistician's role in the prevention of missing data

Continuous(ly) missing outcome data in network meta-analysis: A one-stage pattern-mixture model approach

Appropriate handling of aggregate missing outcome data is necessary to minimise bias in the conclusions of systematic reviews. The two-stage pattern-mixture model has been already proposed to address aggregate missing continuous outcome data. While this approach is more proper compared with the exclusion of missing continuous outcome data and simple imputation methods, it does not offer flexible modelling of missing continuous outcome data to investigate their implications on the conclusions thoroughly. Therefore, we propose a one-stage pattern-mixture model approach under the Bayesian framework to address missing continuous outcome data in a network of interventions and gain knowledge about the missingness process in different trials and interventions. We extend the hierarchical network meta-analysis model for one aggregate continuous outcome to incorporate a missingness parameter that measures the departure from the missing at random assumption. We consider various effect size estimates for continuous data, and two informative missingness parameters, the informative missingness difference of means and the informative missingness ratio of means. We incorporate our prior belief about the missingness parameters while allowing for several possibilities of prior structures to account for the fact that the missingness process may differ in the network. The method is exemplified in two networks from published reviews comprising a different amount of missing continuous outcome data.

Estimands: discussion points from the PSI estimands and sensitivity expert group

ICH E9 Statistical Principles for Clinical Trials was issued in 1998. In October 2014, an addendum to ICH E9 was proposed relating to estimands and sensitivity analyses. In preparation for the release of the addendum, Statisticians in the Pharmaceutical Industry held a 1-day expert group meeting in February 2015. Topics debated included definition, development, implementation, education and communication challenges associated with estimands and sensitivity analyses. The topic of estimands is an important and relatively new one in clinical development. A clear message from the meeting was that estimands bridge the gap between study objectives and statistical methods. When defining estimands, an iterative process linking trial objectives, estimands, trial design, statistical and sensitivity analysis needs to be established. Each objective should have at least one distinct estimand, supported by sensitivity analyses. Because clinical trials are multi-faceted and expensive, it is unrealistic to restrict a study to a single objective and associated estimand. The actual set of estimands and sensitivity analyses for a study will depend on the study objectives, the disease setting and the needs of the various stakeholders. Copyright © 2016 John Wiley & Sons, Ltd.

Missing CD4+ cell response in randomized clinical trials of maraviroc and dolutegravir

BACKGROUND

Missing data can compromise inferences from clinical trials, yet the topic has received little attention in the clinical trial community. Shortcomings in commonly used methods used to analyze studies with missing data (complete case, last- or baseline-observation carried forward) have been highlighted in a recent Food and Drug Administration-sponsored report. This report recommends how to mitigate the issues associated with missing data. We present an example of the proposed concepts using data from recent clinical trials.

METHODS

CD4+ cell count data from the previously reported SINGLE and MOTIVATE studies of dolutegravir and maraviroc were analyzed using a variety of statistical methods to explore the impact of missing data. Four methodologies were used: complete case analysis, simple imputation, mixed models for repeated measures, and multiple imputation. We compared the sensitivity of conclusions to the volume of missing data and to the assumptions underpinning each method.

RESULTS

Rates of missing data were greater in the MOTIVATE studies (35%-68% premature withdrawal) than in SINGLE (12%-20%). The sensitivity of results to assumptions about missing data was related to volume of missing data. Estimates of treatment differences by various analysis methods ranged across a 61 cells/mm3 window in MOTIVATE and a 22 cells/mm3 window in SINGLE.

CONCLUSIONS

Where missing data are anticipated, analyses require robust statistical and clinical debate of the necessary but unverifiable underlying statistical assumptions. Multiple imputation makes these assumptions transparent, can accommodate a broad range of scenarios, and is a natural analysis for clinical trials in HIV with missing data.

Missing data sensitivity analysis for recurrent event data using controlled imputation

Statistical analyses of recurrent event data have typically been based on the missing at random assumption. One implication of this is that, if data are collected only when patients are on their randomized treatment, the resulting de jure estimator of treatment effect corresponds to the situation in which the patients adhere to this regime throughout the study. For confirmatory analysis of clinical trials, sensitivity analyses are required to investigate alternative de facto estimands that depart from this assumption. Recent publications have described the use of multiple imputation methods based on pattern mixture models for continuous outcomes, where imputation for the missing data for one treatment arm (e.g. the active arm) is based on the statistical behaviour of outcomes in another arm (e.g. the placebo arm). This has been referred to as controlled imputation or reference-based imputation. In this paper, we use the negative multinomial distribution to apply this approach to analyses of recurrent events and other similar outcomes. The methods are illustrated by a trial in severe asthma where the primary endpoint was rate of exacerbations and the primary analysis was based on the negative binomial model.